Deep imaging reveals dynamics and signaling in one-to-one pollen tube guidance

Under our experimental conditions, the number of ovules in a flower of the wild-type (WT) Col-0 ecotype of A. thaliana was 46.8 ± 3.3 (mean ± s.d.; n = 10 pistils). The number of pollen tubes in a maximum pollinated pistil was counted 18 h after pollination (HAP) when all three steps of guidance (emergence from the TT, growth on the funiculus, and penetration of the micropyle) had been completed in almost all ovules. NaOH-cleared pollinated pistils stained with aniline blue solution was used to count the number of pollen tubes growing in the TT, which was 70 ± 3.6 (mean ± s.d.; n = 3 pistils; Fig. EV1B−F). Thus, the number of pollen tubes per ovule is approximately 1.5, indicating that almost all ovules must attract a pollen tube on a one-to-one basis for all ovules to be fertilized. In the NaOH-cleared pistil, the pollen tubes within the TT at maximum pollination looked like a pollen tube “bundle” (Fig. EV1B). Some pollen tubes were shown to grow in the z-direction from such a pollen tube bundle, which was thought to be an attracted pollen tubes to the ovule (Fig. EV1C). However, the process of pollen tube guidance on each pollen tube, especially the temporal sequence, was difficult to analyze in the fixed tissue (Fig. EV1B). Thus, we developed a two-photon live imaging method named, “single-locule method” for deep imaging of pollen tube dynamics within a live Arabidopsis ovary expressing fluorescent reporter (Fig. EV2). The pollen tube bundle that entered from style into the ovary TT reached the base of the WT ovary at 16 HAP in the WT ovary (Movie EV1A; Fig. 1A–C). The pollen tube growth rate was not constant and gradually decreased to 1.3 ± 0.1 μm/min (mean ± s.d.; n = 7 pollen tubes) between 4 and 16 HAP (Fig. EV3A). In the yz-projection image, some pollen tubes abruptly separated from pollen tube bundle to pass the septum epidermal (SE) layer, adjusting their growth towards the locule (Fig. 1C). Such directional changes were assumed to indicate pollen tube emergence followed by funicular guidance, as the fluorescent signals of these tubes showed that they elongated along the funiculus after passing through the septum epidermis (SE) (Fig. EV2C), being attracted to the fluorescent-labeled synergid cells (Fig. 1D–G; Movies EV1B,C). Time-lapse imaging of the area encompassing 512 μm from the upper region of the ovary of the limited or maximum pollinated pistils was performed using the single-locule method. In the 314 WT pistils for 1 to 18 h after pollination, pollen tube growth in the TT was observed in 218, and pollen tube emergence and/or funicular guidance was observed in 110 of the pistils with 448 ovules. The two-photon live imaging separately analyzed the processes of pollen tube elongation, pollen tube emergence, and funicular guidance in a living ovary. To the best of our knowledge, this is the first report of such a dynamic recording of the entire guidance of a single pollen tube. Therefore, our single-locule method enabled real-time analysis of pollen tube guidance until it reached the micropyle.

In A. thaliana, it is unclear which ovule would preferentially attract a pollen tube among ovules (Crawford et al, 2007; Hülskamp et al, 1995). As the ovules are linearly aligned and pollen tubes elongate from the top, we hypothesized that pollen tube guidance begins at the apical ovules. To confirm this hypothesis, NaOH-cleared pollinated pistils were observed (Fig. 2). Under maximum pollination, pollen tubes were distributed throughout the entire TT (Fig. EV1B–F). As expected, under maximum pollination, the median distribution of the targeted ovules was 22.0% (205 ovules in 12 pistils, Fig. 2D), which were distributed towards the apex. However, when single pollen pollination was performed (limited pollination, Fig. 2A,B), such apical preferentiality was lost. The single seed distribution in the silique indicated that the germinated single pollen tube had reached it (Fig. 2C). The apical 10% of the ovules were not targeted in both center and side pollination conditions (Fig. 2E,F). The median single seed distribution from the top of the silique was 49% for central pollination (n = 52 siliques, Fig. 2E) and 46% for side pollination (n = 44 siliques, Fig. 2F), both of which were more located at the bottom of the silique than the maximum pollination. No difference in targeting preferentiality was observed between the central and side pollinations, and the apical 10% of ovules were not targeted in either condition (Fig. 2E,F). Therefore, preferential ovule targeting depended on the number of germinated pollen grains, but not on the position of the pollen grain on the stigma. Moreover, the results suggest that a pollen tube can pass the apical ovules, implying that an ovular attraction signal may only be effective locally. In other words, pollen tubes may need to reach the zone of ovular attraction in the TT to be attracted to an ovule. Therefore, we assumed that growing pollen tubes emerge upon receiving some local signals, and next, accordingly analyzed the position of the pollen tube and the time of pollen tube emergence by two-photon live imaging.

Two-photon live imaging showed that pollen tubes are allowed to grow freely by changing available radial positions in the TT under limited pollination (Fig. 3A–D). Pollen tube behavior and growth path were investigated using an SE reporter (Ueda et al, 2017) to determine which pollen tube emerged from the TT (Fig. EV2B,C). Under limited pollination, some pollen tubes gradually moved toward the SE and emerged from the TT, whereas other pollen tubes grew down the TT (Fig. 3A,B). Such gradual movement of the pollen tubes toward the SE and then elongates along the SE within the TT, was observed in many pistils examined, even in the presence of multiple pollen tubes in the TT (white arrow in Fig. 3D, cyan, green, and orange arrows in Fig. 4B; see also Movie EV2). When two pollen tubes elongated in parallel, the one closer to the SE was more likely to emerge (Fig. 3C,D). Therefore, we concluded that pollen tubes were attracted from the top down under conditions of maximum pollination but not limited pollination, for two reasons: pollen tubes adjacent to the SE were selected for emergence in the TT, and under maximum pollination, pollen tubes were more likely to grow adjacent to the SE. These results suggest that pollen tube emergence is stochastically determined depending on the growth position in the TT; the closer pollen tubes grow to the SE, the more pollen tubes are attracted and emerge by sensing some local signals adjacent to the SE.

To assess the properties of the emergence signal(s) adjacent to the SE, we investigated the temporal regulation of pollen tubes that had emerged. As mentioned above, two-photon live imaging of the 110 WT pistils for 1 to 18 HAP revealed 448 ovules with pollen tube attraction. Of these, nine pollen tubes in seven pistils were able to be analyzed the growth rate of pollen tubes within the TT from more than 2.5 h before pollen tube emergence (Fig. 4A). The growth rate of emerged pollen tubes gradually decreased to 0.5 μm/min ~2 h before their emergence (Fig. 4A), while non-emerged tubes maintained a growth rate of 1.4 ± 0.3 μm/min (mean ± s.d.; n = 69 pollen tubes in ten pistils). The xy- and yz-projection images revealed that the pollen tube growth rate decreased following its attachment to the SE within the TT (Fig. 4B,C; EV3B). The median time from pollen tube SE attachment until emergence was 120 min (n = 11; range 45–285 min), consistent with the decreased growth rate. The growth rates of the emerged nine pollen tubes on both septum and funiculus after pollen tube emergence were measured from yz-projection images. The growth rates on the septum and funiculus were 0.5 ± 0.06 μm/min and 0.6 ± 1.6 μm/min, respectively (mean ± s.d.; n = 9 pollen tubes). This result suggests that SE attachment within the TT triggers a reduction in the pollen tube growth rate, persisting throughout its emergence until growing on the septum. However, subsequent funicular guidance showed variation in pollen tube growth rates on the funiculus (0.6 ± 1.6 μm/min). These results suggest that different factors affect pollen tube growth on the septum and on the funiculus.

Ovular cells, especially the two synergid cells, are indispensable for ovule targeting (Higashiyama and Takeuchi, 2015; Lopes et al, 2019). Moreover, soluble secreted materials from the mature ovule act as long-distance pollen tube guidance signals in Torenia (Horade et al, 2013). To investigate the effect of ovule maturity on the pollen tube emergence from the TT, ovule development was examined in ClearSee-treated cleared pistils (Kurihara et al, 2015; Mizuta and Tsuda, 2018). Ovule maturation was indicated with the appearance of a fluorescent signal of the FGR8.0 reporter, which combines ovular cell fluorescent markers (Völz et al, 2013). Maturation commenced in the middle to lower portion of the ovary at stage 11, then extended to both the upper and lower extremities until stage 14 (Fig. EV4A). Even on the day of flowering, at stage 14, some ovules in the topmost ovary remained immature (arrows in Fig. EV4A). This result correlated with the ovule targeting preference under limited pollination conditions (Figs. 2 and 3), suggesting that ovule maturation contributes to ovule targeting. It was also consistent with our hypothesis that the emergence signal(s) are derived from the ovules and reach to the inner surface of the TT.

The live imaging results suggest that pollen tubes in the TT are guided before they emerge, and that ovule maturation may be involved in pollen tube emergence. However, long-range ovular guidance has been a subject of debate (Higashiyama and Takeuchi, 2015). The female gametophyte and its synergid cell are the source of the ovular attraction signal (Takeuchi, 2021). The central cell has also been reported to be involved in fertilization recovery when the first tube fails fertilization (Meng et al, 2023). Due to the observed correlation between ovule maturation and pollen tube targeting under limited pollination conditions as described above, we hypothesized that the female gametophyte might be the origin of the long-distance emergence signal that reaches the inner surface of the septum. Thus, ClearSee-treated cleared pistils of the WT and three ovular mutants were examined. The ovular tissues of each sample are as follows; (1) WT has both sporophytic and gametophytic tissues in an ovule, (2) In the aintegumenta (ant) mutant ovule, both sporophyte and gametophyte are severely defective, due to a complete loss of megasporogenesis (Elliott et al, 1996), (3) In the determinant infertile 1 (dif1) mutant ovule, it lacks a female gametophyte due to meiotic chromosome missegregation (Bhatt et al, 1999) even though the sporophytic tissues are present; (4) In the inner no outer (ino) mutant ovule, it lacks sporophytic outer integuments of the ovule (Villanueva et al, 1999). In each of these pistils, the number of emerged pollen tubes and that of ovules showing funicular guidance and polytubey were examined (Fig. 5). Pollen tubes elongating along the female tissues including septum surface and funiculus were defined as adhesions (Lord, 2000). Ovule showing multiple pollen tubes on the funiculus was also defined as polytubey (Fig. 5). (1) In the WT, pollen tubes emerging from the TT adhered to the septum surface, and then elongated along the funiculus toward the ovule (Fig. 5A). In the pistil at 6 HAP, 20.7 ± 4.7 pollen tubes emerged from the TT, and 17.0 ± 5.6 ovules showed funicular guidance, of which 0.3 ± 0.6 ovules showed polytubey as multiple pollen tubes on the funiculus (mean ± s.d.; n = 9 pollen tubes; Fig. 5B–D). To confirm the requirement of sporophytic/gametophytic tissues in the ovules for pollen tube behaviors, three mutants was similarly examined. (2) In ant mutant ovaries, there was a similar number of emerged pollen tubes (15.0 ± 2.6 ovules, mean ± s.d., ns; Fig. 5B) but a significantly reduced number of ovules showing funicular guidance (0.7 ± 0.6 ovules, mean ± s.d., ***p < 0.001; Fig. 5C) compared to the WT. (3) In a female gametophytic mutant dif1, although micropylar guidance by the ovular gametophytic cells was completely impaired, significant increase in the number of pollen tubes which emerged from the TT was observed compared to those from WT (35.3 ± 6.7 pollen tubes, mean ± s.d., *p < 0.05; Fig. 5B), whereafter many of them lost their way to the micropyle and continued their disordered growth (Fig. 5A). As a results, polytubey was significantly increased in a dif1 ovary (3.0 ± 1.0 ovules, mean ± s.d., **p < 0.01; Fig. 5A,D). These observations suggest that ovular sporophytic cells positively regulate pollen tube emergence, whereas ovular gametophytic cells negatively regulate pollen tube emergence. Thus, we also investigated the ovular sporophytic mutant, the ino mutant. (4) In ino mutant ovaries, there was a similar number of emerged pollen tubes (10.3 ± 4.9 pollen tubes, mean ± s.d., ns; Fig. 5B), whereas a significantly reduced number of ovules showing funicular guidance (0.7 ± 0.6 ovules, mean ± s.d., ***p < 0.001; Fig. 5A,C) and that of pollen tubes attached to the septum and funiculus (Fig. 5A), suggesting that the sporophytic ovular outer integument enhances pollen tube adhesion to the maternal tissue surface. Consistently, disordered pollen tube adhesion to the septum and funicular surfaces was observed in the (2) ant mutant that also lacks outer integuments (Fig. 5A). Thus, we concluded that sporophytic ovular outer integument enhances pollen tube adhesion to the maternal tissue surface including septum and funiculus. The results of pollen tube emergence in the (3) dif1 and (4) ino also suggest that the female gametophytic cell-derived factors negatively regulate pollen tube emergence and positively regulate polytubey block on the funiculus, independent of fertilization.

To investigate the regulation by the outer integument, pollen tube emergence in the ino ovary was assessed through two-photon live imaging. In the WT, pollen tubes gradually changed their growth direction within the TT, then elongated along the SE for ~50–300 μm before emergence (Fig. 4E). In contrast, pollen tubes in the ino TT did not grow along the SE (Fig. 4E) and emerged rapidly without decreasing their growth rate (Fig. 4A,C). After emergence from the TT, it was observed that the pollen tube in the locule elongated upward with disordered growth without growing toward the ovule (Fig. 4D). The Arabidopsis TT comprises longitudinally elongated cells and a large number of extracellular matrix components (Crawford et al, 2007); however, there were no obvious morphological differences between the TTs of WT and ino (Fig. EV4B). These results demonstrate that an unknown long-distance guidance signal(s) derived from the ovular sporophytic outer integument affects the pollen tube within the TT, which facilitates pollen tube emergence and attachment to the female tissue surfaces.

We investigated when and where the female sporophytic or gametophytic cells regulate pollen tube emergence from the transmitting tract and polytubey block on the funiculus. As mentioned above, FER is widely expressed in female tissues, plays important roles, and is involved in polytubey block at both the septum and funiculus (Duan et al, 2020; Zhong et al, 2022). The number of ovules showing single pollen tube or multiple pollen tubes (polytubey) on the funiculus at 4, 6, and 18 HAP was analyzed by NaOH-cleared WT and fer pistils stained with aniline blue solution. We also analyzed the ovules accepting homozygous generative cell-specific 1 (gcs1) mutant pollen tubes as controls because polytubey in gcs1/hapless 2 (hap2) pollen-pollinated WT pistil occurs by a different mechanism, known as fertilization recovery (Beale et al, 2012; Kasahara et al, 2012; Nagahara et al, 2015; Duan et al, 2020). In the fer mutants, one-to-one pollen tube guidance was impaired, resulting in multiple pollen tubes on the funiculus of each ovule (polytubey, Fig. 6A). It should be noted that most pollen tubes on the fer funiculus did not overlap as reported in the magatama mutant (Shimizu and Okada, 2000), but most gcs1 pollen tubes were attracted to the same route as the first pollen tube on the funiculus, as previously reported (Kasahara et al, 2012), consistent with different mechanisms underlying the polytubey of these mutants (Fig. 6A).

We first hypothesized that FER-dependent blocking (Zhong et al, 2022) mainly regulates one-to-one pollen tube guidance at the septum. To investigate blocks during pollen tube emergence, WT and fer mutant pistils at 4 HAP were investigated when fertilization was incomplete in most ovules. The number of emerged pollen tubes was significantly increased in the fer pistil, even at 4 HAP, but not in the WT pistil (Fig. 6B). However, the position of pollen tube emergence on the septum surface (apical preference) did not differ significantly (Fig. 6C). This finding supports the notion that FER-dependent sporophytic signaling functions upon emergence onto the septum as a block by restricting the number of emerged pollen tubes.

We further investigated polytubey blocks at funicular guidance after pollen tube emergence. The distribution of ovules with multiple pollen tubes on the funiculus, defined as polytubey, was investigated in NaOH-cleared pistils. Polytubey was scarce among WT ovules but prevalent in fer, even at 4 HAP (Fig. 6D). We also analyzed the mutant of LORELEI (LRE) that interacts with FER in synergid cells but is not expressed in sporophytic tissues (Liu et al, 2016). In the lre mutant, both the number and the position of emerged pollen tubes were not significantly different from those in the WT (Fig. 6B,C), whereas polytubey was observed at high frequency at 4 HAP as well as in the fer mutant (Fig. 6A,D). The gcs1 mutant and WT showed few polytubey events at 4 and 6 HAP (Fig. 6A,D). This result suggests that FER mainly controls the number of emerged pollen tubes at the septum, whereas synergid cell-dependent gametophytic signal(s) regulate the number of pollen tubes to enter into the funiculus. These multistep controls achieve one-to-one pollen tube guidance as polytubey blocks.

A mutation in synergid cell-specific LRE caused polytubey at the step of funiculus entry step (Fig. 6A) but not caused increasing emerged pollen tubes (Fig. 6B). We hypothesized that the gametophytic FER- and LRE-dependent signaling regulates polytubey block on the funiculus, even before the arrival of the first pollen tube at the synergid cell that exerts the blocking signal (Duan et al, 2020). The spatiotemporal behavior of pollen tubes during polytubey was observed under the single-locule method. Live imaging results show that across 110 WT pistils from 1 to 18 HAP, 448 ovules displayed pollen tube attraction of which 18 possessed multiple pollen tubes on the funiculus, which was defined as polytubey, representing a rate of 4.0% (Fig. 7A; Movie EV3A). This ratio is comparable to that reported for polytubey in NaOH-cleared pistil (around 10% in Zhong et al, 2022; 1.6% in Capron et al, 2008). Our live imaging showed more polytubey in the fer (27 pistils: 111 ovules exhibiting attraction, 37 with polytubey; a rate of 33%) than in the WT (Fig. 7B,D), similar to the previous report in the NaOH-cleared pistil (over 40%, Zhong et al, 2022), and more than that in the heterozygous fer mutant (fer/+) NaOH-cleared pistil (about 10%, Huck et al, 2003). Polytubey was also frequently observed in lre under the single-locule method (19 pistils: 71 ovules displaying attraction, 35 with polytubey; a rate of 49%, Fig. 7C,D), which was more than the previous report in the NaOH-cleared pistils (17.9%, Tsukamoto et al, 2010).

Next, the timing of funicular guidance initiation between the first and second pollen tubes was also analyzed. All ten polytubey events observed in the WT (in which the second pollen tube was attracted without repulsion) were observed within 45 min (Fig. 7D). Similarly, 31 of 37 (84%) and 29 of 35 (83%) polytubey events were observed within 45 min for lre and fer, respectively. These results suggest that the polytubey block is not strictly prohibitive within 45 min after entry of the first pollen tube onto the funiculus, even in the WT. The frequent polytubey within 45 min in both fer and lre suggests that the distantly locating synergid cell is involved in rapid but non-strict blockage of entry onto the funiculus. To determine whether such blockage depends on the sporophyte or gametophyte, the number of polytubey was analyzed in the NaOH-cleared heterozygous fer and lre mutant pistils. Heterozygous fer and lre mutants showed frequent and comparable polytubey on the funiculus at 6 HAP, suggesting that it is not the sporophytic but rather the gametophytic signal that is required for blocking second pollen tube entry onto the funiculus (Fig. 7E). Notably, we observed that several pollen tubes returned from the funiculus to the septum (i.e., repulsion) in both fer and lre mutants even they were the first pollen tubes, but not in the WT (arrows in Fig. 7B,C; Movies EV3B,C). Therefore, this new repulsive signal may be inhibiting the funicular guidance of excess pollen tubes, which is still active in both the unfertilized fer and lre ovaries. There are at least two stages of polytubey blockage after the start of funicular guidance: The first stage is incomplete, and begins from the time when the first pollen tube begins funicular guidance, but stronger blockage works after 45 min as the second stage. Under our observation, the first block was defective in both the fer and lre mutants, but the second blocking mechanism was evident in both mutants, albeit incompletely. Our study reveals a repulsion mechanism at the pollen tube entry step on the funiculus, suggesting an important new aspect in the multiple polytubey blocks.

Arabidopsis ecotype Columbia (Col-0) was used as the WT plant. LAT52p::mTFP1, sGFP, Venus, TagRFP, and mApple (Mizuta et al, 2015), FGR8.0 (Völz et al, 2013), MYB98p::GFP (Kasahara et al, 2005), and HDG11p::NLS-YFP (Ueda et al, 2017) have been previously described. Nuclei in the septal epidermal cells of HDG11p::NLS-YFP are labeled with YFP. Homozygous mutant seeds of ant (SALK_022770), dif1 (rec8-1; SALK_091193), fer (fer-4; GK-106A06) (Haruta et al. 2014), gcs1 (SALK_135496), ino (ino-4; N6148), and lre (lre-5; CS66102), and heterozygous seeds of fer (fer-4; GK-106A06) and lre (lre-5; CS66102) were obtained from the Arabidopsis Biological Resource Center at Ohio State University (Columbus, OH, USA) or the GABI-KAT line (Rosso et al, 2003). Seeds were germinated on agarose plates at 22 °C under 24 h light. The primers used for genotyping each mutant are listed in Table EV1. Fourteen-day-old seedlings were transferred into a mixture of vermiculite and potting compost. Seedlings were grown at 21–24 °C under long-day conditions (16 h light/8 h dark).

The single-locule method was developed for live imaging of pollen tube guidance within a living pistil (Fig. EV2). This method normalizes the optics for Arabidopsis two-photon live imaging (Mizuta et al, 2015) based on a previous method (Rotman et al, 2003). Fully mature stage 14 flowers were used to avoid potential pistils with immature ovules. The flowers were emasculated 18–24 h before being hand-pollinated. A frame made of a 0.2 mm thick silicone film (AsOne, Osaka, Japan) was placed on the glass bottom dish as a mold (Fig. EV2A, Setting), as previously described (Mizuta and Higashiyama, 2014). Pollen germination medium (PGM) (Boavida and McCormick, 2007) and low-melting-point agarose (NuSieve GTG agarose; Lonza Group Ltd., Basel, Switzerland) were poured into the mold. Immediately after hand pollination, a pistil was collected on a glass slide with double-sided tape. Pistils were cut open using an ophthalmic knife (MANI, Inc., Tochigi, Japan) at the valve margins (Fig. EV2A, Dissection). The remaining locule was immediately placed horizontally on the cut side (Fig. EV2A, Setting), and the flower stalk was embedded in the solid PGM (Fig. EV2A, Imaging). Liquid exuding from the solid PGM filled the bottom of the locule by capillary action. A piece of 0.2 mm thick silicone film was then placed on top of the remaining locule as a weight to prevent lifting. The silicone frame was covered with a 0.1 mm thick silicone film to maintain high humidity. Wet paper wipes were placed in a glass bottom dish to maintain humidity (Fig. EV2A, Setting), and the system was sealed with Parafilm (Bemis Flexible Packaging, Oshkosh, WI, USA).

The imaging system used was based on a previous study (Mizuta et al, 2015). Images were acquired using a laser-scanning inverted microscope (A1R MP; Nikon, Tokyo, Japan) equipped with a 25× water immersion objective lens (CFI 75 Apo 25xW MP, WD = 2 mm; NA = 1.10). A handmade water supply system was equipped with an objective lens, which enabled long-term time-lapse imaging. Emitted fluorescence signals were detected using non-descanned GaAsP PMT detectors (Nikon). To reduce plant autofluorescence and photodamage, excitation wavelengths of 980–1000 nm were used for live imaging and 850 or 860 nm for imaging of fixed samples (Mizuta et al, 2015). Time-lapse imaging was initiated 1–2 HAP at 10- or 15-min intervals. Z-stack images were taken using multiple z-planes at 1–15-μm intervals for 3D construction, projection images, and optical sections inside the ovary. To reconstruct a large image, image stitching was automatically performed using NIS-Elements v4.10 software (Nikon). All images were analyzed with Adobe Photoshop (Adobe Systems, Inc., San Jose, CA, USA) and Fiji (Schindelin et al, 2012) software.

Pollen tube growth rates and distribution were calculated by manually tracking the pollen tube tips on the xy-projected images using Fiji software. As the pistil grew under the single-locule method, an average of three random points on the SE were used to normalize the xy-movement associated with pistil development.

Pistil clearing was performed using the ClearSee.v2 reagent as previously described (Mizuta et al, 2016; Mizuta and Tsuda, 2018). ClearSee.v2 achieves tissue transparency within the remaining spatial arrangement of organs and fluorescent signals in the Arabidopsis pistil (Kurihara et al, 2015). Pistils were collected and fixed with 4% paraformaldehyde (w/v) for 1 h at each time point and then cleared in a ClearSee.v2 solution for 4 weeks, according to previous reports (Mizuta and Tsuda, 2018). To assess ovule maturation, FGR8.0 immature pistils were fixed before flowering at stages 10, 11, and 12 (Smyth et al, 1990) and then cleared with ClearSee.v2. Pistils at stage 14 were emasculated at stage 12. To observe ovary tissue structure, pistils were stained with 0.1% (w/v) calcofluor staining solution after clearing with ClearSee.v2. Transparent pistils were observed using two-photon excitation microscopy (2PEM) (A1R MP, Nikon) at a suitable excitation wavelength for each fluorescent protein (Mizuta et al, 2015).

Stage 12 buds of WT flowers were emasculated and used for experimentation after being allowed to mature for 24 h. For hand pollination, single pollen grains were applied with tweezers to the center or side of a stigma under a stereomicroscope. After 3 d, 52 pistils with central pollination and 44 pistils with side pollination were collected. Sample fixation and clearing by ClearSee.v2 were performed in the same way as described above. After one week of clearing, transparent siliques were observed under a stereomicroscope (SZX7; Olympus, Tokyo, Japan) with an i-NTER LENS optical adapter (Microscope Network Co. Ltd.). The seed-set distribution within the silique was analyzed using Fiji software to determine the percentile of seeds present in each of the ten percentiles of ovary length, from the most apical (1–10%) to the most basal (90–100%) aspect of each ovary.

Stage 12 buds of WT flowers were emasculated and used for experimentation after being allowed to mature for 24 h. To achieve maximum pollination, stamens from the opening flower were pinched with tweezers and gently brushed across the emasculated stigma to attach pollen under a stereomicroscope. The 12 pollinated pistils were collected 6 HAP. Sample fixation and clearing by ClearSee.v2 were performed as described above. An ovule with at least one pollen tube on its funiculus was defined as the target ovule. Fiji software was used to measure the distribution of the funiculus of the targeted ovule. The percentile of seeds present in each of the ten percentiles of ovary length was determined, from the most apical (1–10%) to the most basal (90–100%) aspect of each ovary. The distribution of 205 ovules and the tips of the most growing pollen tubes in the 12 pistils were analyzed.

Pistils were emasculated 18–24 h prior to hand pollination. Pistils from WT, homozygous fer (–/–), heterozygous fer (+/–), homozygous lre (–/–), and heterozygous lre (+/–) plants were pollinated with WT pollen. The pollen of the homozygous gcs1 mutant was pollinated with WT pistils. Pistils collected 4, 6, and 18 h after pollination (HAP) were fixed with a 9:1 mixture of ethanol:acetic acid and then incubated in 1 N sodium hydroxide (NaOH) for clearing (Kasahara et al, 2012). After overnight incubation, pistils were stained with 0.1% (w/v) aniline blue in a K₃PO₄ buffer for more than 10 min (Kobayasi and Atsuta, 2010). The stained pistils were rinsed with sterilized distilled water and then observed using two-photon excitation microscopy (2PEM) at an excitation wavelength of 850 nm (A1R MP, Nikon). The distributions of the ovules showing either single pollen tube attraction or polytubey were analyzed manually with Fiji software.

The visualization and statistical analyses of data were performed using GraphPad Prism v9.0 (GraphPad Software, San Diego, CA, USA). Data were analyzed using one-way analysis of variance (ANOVA), followed by Dunnett’s multiple comparisons tests or Tukey’s multiple comparisons tests to analyze the difference in the number of emerged pollen tubes, funicular guidance, ovules with polytubey, and the location of pollen tube emergence.

The WT pistil of A. thaliana was cut open using an ophthalmic knife (MANI, Inc., Tochigi, Japan) at the valve margins to expose the inner ovules. The pistil was observed with a scanning electron microscope (VHX-D500; Keyence, Osaka, Japan).